Pull the numbers yourself from Seller Central rather than from their deck. Compare contribution profit, organic share of sales, and advertising efficiency across matched windows before and after they started. Then ask who inside their company touched the account, because work credited to a subcontractor is not their work.
The short version
- Your data, your report. Any audit built from the provider's own slides is a review of their storytelling.
- Match the windows. Same length, same season, same promotional events, or the comparison means nothing.
- Contribution profit is the headline. Revenue growth funded by ad spend is not growth.
- Organic share is the durability test. A brand that only moves when the ads run has not been built.
- Ask who did the work. Named people on their payroll, or you cannot hold anyone to the result.
The mistake almost every seller makes
The common failure is auditing on the wrong cadence. Sellers review an agency at renewal, using the deck the agency prepared, comparing whatever period makes the trend look best. By then the money is spent and the decision is emotional, because there is a relationship and there are people you like.
Do it quarterly instead, from your own export, on numbers you chose in advance. It takes an afternoon and it is the difference between managing a supplier and hoping.
The five numbers to pull
- Contribution profit. Revenue minus landed cost, Amazon fees, and total ad spend. This is the only number that answers whether the relationship is worth its own cost.
- Organic share of sales. Units not attributed to advertising, divided by total units. Rising share means the position is being built. Falling share while revenue rises means you are renting sales.
- Total advertising cost of sale across the account. Not campaign-level highlights. The blended number, including everything that ran.
- Conversion rate by listing. Where listing and creative work either happened or did not.
- Return rate and rating trend. The two that quietly destroy a product while the revenue chart still looks fine.
Then set them beside the fee. A provider costing $2,400 a month has to produce more than $28,800 a year of contribution profit that would not otherwise exist. That is not a hostile standard, it is arithmetic, and any provider should be willing to be measured against it.
Failure modes, ranked by what they cost you
| What you find | What it means | Cost |
|---|---|---|
| Revenue up, contribution profit flat | Growth bought with ad spend | Highest, and it compounds monthly |
| Organic share falling | No ranking or listing work underneath the ads | High, and it shows up when you pause spend |
| Reporting on impressions and clicks | Metrics chosen because they always rise | Moderate, mostly wasted attention |
| Change log thinner than the report | Nobody is in the account | Moderate to high, depending on how long |
| Work traced to a subcontractor | No accountability, variable quality | Unpredictable and hard to fix |
The subcontracting check
This is the one people forget to run. Ask for the names and roles of everyone who touched your account last month, and where they sit. A provider doing the work will answer in a sentence. A provider forwarding it elsewhere will talk about partners and networks.
I have a bias here worth declaring. We do not subcontract anything. Our creative studio is in Dubai, our sourcing studio is in Guangzhou, and every operator is on our payroll, because when quality slips I need to be able to find the person and fix it. You do not have to hire that model, but you should know which model you have bought.
The change log check
Ask for the actual list of changes made in the account over 30 days: campaigns created or paused, listing edits, images replaced, keywords added, price changes. Compare it to the length of the monthly report. When the report is longer than the log, you are paying for reporting.
What most agencies will not tell you
The reporting period is chosen after the results are known. Almost nobody picks the comparison window in advance, so a soft month gets compared against a softer one and a strong quarter becomes the baseline for everything. Fix this by defining the windows and the five numbers on day one, in writing, before anyone has an incentive.
Second, attribution is hard and most decks pretend it is not. If a competitor went out of stock, if Amazon changed a browse node, if your category had a seasonal spike, some of the improvement was not the agency. A provider who volunteers that context is more trustworthy than one who claims every good month. Ask directly what happened this quarter that had nothing to do with them.
Related answers
- KPIs an Amazon agency should report weekly
- What does a good Amazon account audit include
- Amazon agency red flags to watch out for
- Monthly cost for outsourced Amazon management
- Amazon agency pricing and economics: the complete guide
If you want a second opinion on what your current provider produced, the free audit is at Flapen.
